At present, database is mainly applied in Online Transaction Processing (OLTP) and Decision Support System (DSS).
OLTP is a kind of operational processing for particular applications of enterprises. Generally, OLTP includes transactional database applications of daily operational tasks in enterprise management, such as timely registration of a new record, and retrieval and modification of a record or a set of records, which generally doesn't need complicated retrieval or analysis. Real-time performance, and security and completion of data are the most concerned parts in OLTP. OLTP is generally used by business staff, so when an OLTP is designed, corresponding transactional databases are generally designed from the view of the business staff. That is, meanings in practical business logics are more considered for the fields in the transactional databases than questions about computer technology.
DSS is a kind of system providing an analysis processing on a basis of a large amount of historical data, for managers' decision analysis. Existing DSS is generally realized based on cooperation of various technologies, such as Data Warehouse (DW), Online Analytical Processing (OLAP), Data Mining (DM), etc. The DW provides a nonvolatile (i e, unmodifiable), subject-oriented and compositive data set which varies with time, to support decisions made by managers. Data is extracted, transformed and loaded from a transactional database to the DW. DW is generally designed from the view of the subject to be analyzed by a skilled person. A subject is generally related to multiple fields in the transactional database.
Due to remarkable differences between a transactional database and a DW, it is very difficult for a skilled person to design a mapping between the transactional database and DW. The business logic contained in the transactional database is transparent in semantic for the DW and the transactional database is developed based on requirements of business staff, however, which may not satisfy development requirements of skilled persons.
In conventional solutions, a mapping from a transactional database to a DW needs to be designed manually, which is complicated and time-consuming, costs high in development and maintenance and has poor transferability and universality.
In conventional solutions, Model Driven Architecture (MDA), such as Eclipse Modeling Framework (EMF), has been developed to design a transactional database model. Besides, conceptual designs for automatically designing a DW model have appeared, such as E/R models proposed by Hüsemann, Lechtenbörger and Vossen in 2000 or proposed by Phipps and Davis in 2002. Nevertheless, these solutions both fail to solve technology problems in designing a mapping from a transactional database to a DW effectively.
More relative information can be found in a Chinese patent publication No. CN101286151A.